Anthropic AI unemployment analysis challenges Amodei’s job-loss warnings

A new Anthropic AI unemployment analysis from the company’s head of economics says artificial intelligence has not produced a broad rise in U.S. joblessness so far. The finding matters because it sits uneasily beside CEO Dario Amodei’s repeated warnings that AI could soon cause severe losses in white-collar work.

Peter McCrory, Anthropic’s head of economics, wrote in a lengthy essay on X this week that the company’s economic research over the past 18 months does not show a significant AI-driven effect on the U.S. labor market yet. He said he does not expect unemployment to be noticeably higher a year from now because of AI.

Has AI increased unemployment?

McCrory pointed to several labor indicators in arguing that it has not. He said the U.S. unemployment rate was 4.2% in June, a level the Federal Reserve associates with full employment, while job openings were roughly in line with the number of unemployed workers and prime-age employment remained near multi-decade highs.

He also said newer Bureau of Labor Statistics data does not show a relative worsening in unemployment for workers in jobs with many tasks that Claude, Anthropic’s AI assistant, is used to automate. In his account, workers in more AI-exposed roles have not fared worse in unemployment than workers in less exposed occupations.

McCrory’s explanation is that current AI systems have an uneven set of abilities. He cited the “jagged” capability profile associated with Wharton professor Ethan Mollick’s work, saying no job in the Labor Department’s O*NET taxonomy has all of its tasks covered by Claude.

Complex work, McCrory said, still depends on people to guide AI tools and spot mistakes. He said Anthropic’s usage evidence points to Claude often serving as a thought partner rather than a replacement, and that users with more subject expertise tend to do better with the technology.

What did Dario Amodei warn about?

Amodei has made some of the AI industry’s starkest public statements about labor disruption. In May 2025, he told Axios that AI could eliminate half of entry-level white-collar jobs and push unemployment to 10% to 20% within one to five years.

In a January 2026 essay titled “The Adolescence of Technology”, Amodei described AI as a general substitute for human labor and warned it could leave some workers unemployed or stuck in very low-wage jobs, according to Investopedia’s coverage of the essay.

By May, Amodei also discussed a different mechanism: AI could raise output by automating much of a job while leaving people to perform the remaining work. That argument draws on the Jevons paradox, the idea that efficiency can increase total use of a resource or service because it becomes cheaper or more productive.

In June, according to Business Insider, Amodei again warned that large and lasting job losses could be tied to the nature of the technology and called for policy responses, including wage insurance and universal basic income.

Where the two views still overlap

McCrory’s analysis does not say there is no labor-market risk. He noted that hiring has softened over the past year for young workers in highly AI-exposed roles, a pattern also discussed in Stanford research on entry-level jobs.

He also cited Bureau of Labor Statistics projections showing slower growth through 2034 for occupations such as technical writers, data entry workers and customer support representatives. Those are jobs where AI exposure is high and where demand may weaken even if overall unemployment remains low.

The disagreement inside Anthropic is therefore about timing and scale. McCrory’s reading of current data undercuts Amodei’s most urgent scenario, while still leaving room for the view that AI may reshape hiring and make early-career white-collar work more vulnerable.

McCrory also acknowledged that future advances could change the picture. If AI begins automating innovation through recursive self-improvement, he wrote, standard economic models can allow for far more disruptive outcomes, though he does not see that showing up in the near-term labor data.

This story draws on original reporting from Fortune.